1990-1996 Nucleic Acids Research, 1995, Vol. 23, No. 11 ©1995 Oxford University Press

Transformation of Escherichia coli with large DNAmolecules by electroporationYuLing Sheng, Valeria Mancino and Bruce Birren*

Division of Biology 147-75, California Institute of Technology, Pasadena, CA 91125, USA

Received January 24, 1995; Revised and Accepted April 4, 1995

ABSTRACT

We have examined bacterial electroporation with aspecific Interest in the transformation of large DNA, I.e.molecules >100 kb. We have used DNA from bacterialartificial chromosomes (BACs) ranging from 7 to 240kb, as well as BAC ligatlon mixes containing a range ofdifferent sized molecules. The efficiency of electro-poration with large DNA is strongly dependent on thestrain of Escherichia coll used; strains which offercomparable efficiencies for 7 kb molecules differ Intheir uptake of 240 kb DNA by as much as 30-fold. Evenwith a host strain that transforms relatively well withlarge DNA, transformation efficiency drops dramati-cally with Increasing size of the DNA. Molecules of 240kb transform ~30-fold less well, on a molar basis, thanmolecules of 80 kb. Maximum transformation of largeDNA occurs with different voltage gradients and withdifferent time constants than are optimal for smallerDNA. This provides the opportunity to Increase theyield of transformants which have taken up large DNArelative to the number incorporating smaller mol-ecules. We have demonstrated that conditions may beselected which increase the average size of BACclones generated by electroporation and compare theoverall efficiency of each of the conditions tested.

INTRODUCTION

Transformation of bacteria with exogenous DNA has played acentral role in defining the molecular events that underlie geneticphenomena (1). High efficiency transformation makes thecreation of representative libraries feasible when large numbersof colonies are needed or when the source DNA is limited.Although chemical treatments of the cells have historically beenthe means of promoting competence to take up DNA, electro-poration produces the highest efficiencies (reviewed in 2),making it the method of choice for construction of genomiclibraries. Electroporation, in which cells and DNA are subjectedto brief high voltage pulses, is thought to involve the transientappearance of pores (3,4), through which the DNA is driven intothe cell by the electric field (5,6). Many techniques for producingthe electric field have been applied (7-11). The most commonmethod is capacitor discharge, which produces a field that decaysexponentially over time (12).

The absolute efficiency of electroporation, or the number oftransformants per amount of input DNA, is critical for libraryconstruction and many studies have reported procedures forincreasing this efficiency. Efficiency can be influenced by factorssuch as temperature (13) and other electric field parametersselected by the user, including the voltage gradient, resistance andcapacitance (10,14,15). In addition, qualities associated with thehost cells, such as their genetic background (2), growth condi-tions (16) and post-pulse treatment (12), as well as the topologicalform and treatment of the DNA samples (17,18), will also alterthe number of transformants recovered.

We are engaged in the construction of libraries containing largeDNA fragments using an F factor-based vector to create bacterialartificial chromosomes (BACs) (19). In this work, partially-cut,size-selected DNA of several hundred kilobase pairs is ligated toa 7 kb vector and the ligation mix is electroporated to generate alibrary of clones. While a wealth of empirical observations and astrong theoretical framework exists to guide the cloning ofsmaller DNA fragments, the application of these procedures tolarger sized DNA molecules is a recent phenomenon. Conse-quently, little data exist to direct cloning of large fragments inbacteria and it is not clear how well we can predict the optimalconditions for any of the steps in the procedure. For example, ourlack of understanding about the uptake of large DNA moleculesinto bacteria has made it difficult to select conditions for theelectroporation of BAC DNA. We have used the BAC system toinvestigate the process of electroporation, to establish whichparameters are specifically important for the transformation oflarge DNA. BAC clones are well suited for such studies, sincethey are available in a variety of defined lengths, with eachcontaining a common set of vector sequences. In addition, theblue/white color screen provides a rapid indication of the relativetransformation efficiency of large versus small molecules.Finally, the F factor-derived portion of the molecules ensures thatmultiple BACs will not reside in a single clone, simplifying theinterpretation of transformations.

MATERIALS AND METHODS

Plasmid DNA preparation

BAC mini-preps from overnight cultures were performed either inthe Autogen 740 automated mini-prep system (Integrated Separ-ation Systems, Natick, MA) using 3 ml cultures or by hand using1.5 ml cultures. The manual preparations were accomplished using

* To whom correspondence should be addressed at present address: Whitehead Institute/MIT Center for Genome Research, 1 Kendall Square, Building 300,Cambridge, MA 02139, USA

Nucleic Acids Research, 1995, Vol. 23, No. 11 1991

an alkaline lysis procedure (20) with solutions I, II and IE preparedaccording to Sambrook et al. (21). Cell pellets were resuspendedin 100 (il chilled solution I and placed on ice. To this was added 200ul freshly prepared solution II, the tubes were inverted 7-9 timesto mix and replaced on ice. After adding 150 ul solution El, thetubes were again inverted 7-9 times and immediately spun for 6min at full speed in the microfuge at room temperature. Thesupernatant was poured into a new tube and the nucleic acidprecipitated by addition of 1 ml room temperature ethanol. Aftermixing by inversion, samples were immediately spun for 6 min atfull speed in the microfuge at room temperature. Precipitated DNApellets were washed with 1 ml 70% ethanol, air dried andresuspended in 20 ul TE (10 mM Tris, 1 mM EDTA, pH 8.0). DNApellets from the Autogen preps were resuspended in 50 ul. Forquantitarion of B AC DNA, 2 (il mini-prepped DNA was digestedwith Noil (which excises the insert from the vector) and run on amini gel of 0.7% agarose, along with serial dilutions of full-length,linear X DNA and HindUI digests of X DNA. Gels were stained in1 Jig/ml ethidium bromide for 20 min. After photography, theintensities of the BAC samples were compared with those of thestandards. Sample DNA was always bracketed by a range of bothhigher and lower concentrations of the standards and photographicexposure times were selected to avoid saturation. For electropora-tion, samples were diluted in water and stored at 4°C.

Preparation of competent cells

Competent cells were prepared by the procedure supplied with theelectroporarion device (Life Technologies, Gaithersburg, MD), withminor modifications to the protocol. A fresh overnight culture wasdiluted-l:1000for growth in SOB (without Mg2+) (21). Cells weregrown at 37 °C with shaking at 200 r.pjn. to an OD550 of 0.7 (neverhigher than 0.8, except where noted). Cells were harvested byspinning in a GSA rotor for 10 min at 5000 r.p.m. and resuspendedin a volume of cold 10% sterile glycerol equal to the original culturevolume. Cells were collected by spinning for 10 min at 5000 r.p.m.at 4°C. This washing and spinning procedure was then repeated,again using a volume of 10% glycerol equal to the original culturevolume. After decanting the supernatant, cells were resuspended inthe volume of glycerol remaining in the centrifuge bottles (-60 ml)and spun for 10 min at 7000 r.p.m. in an SS34 rotor. After decantingthe supernatant, cells were resuspended in 10% glycerol, using avolume of between 2 and 2.25 mVl initial culture. Cells werealiquoted to microfuge tubes (100-200 ul/tube) and frozen quicklyin a dry ice-ethanol bath. Cells were stored until use at -70°C.Hectrocompetent cells were purchased in the case of DH5a (LifeTechnologies, Gaithersburg, MD) and JS4 (BioRad Laboratories,Richmond, CA).

Electroporation

Electroporaiion was carried out using either a BioRad GenePulser or Gene Pulser II. Cuvettes with a 0.1 cm gap (BioRadLaboratories) were placed on ice to chill. Electrocompetent cellswere thawed and 30 (il added to microfuge tubes placed on ice.Two ul DNA was added and, after mixing, the DNA and cellswere transferred to the cuvette. Electroporation conditions were100 Q, 1.8 kV and 25 uF, except where otherwise noted.Electroporations were performed in duplicate. After electropora-tion, 0.5 ml SOC was immediately added to the cuvette and thecontents were then transferred to sterile glass culture tubes for 45min growth with shaking at 37°C. For each electroporation, 2.5,

25 and 250 ul of cells, each in duplicate, were spread on LB platescontaining 12.5 ug/ml chloramphenicol. For transformations ofligations with the BeloBAC vector, plates contained 50 (ig/mlX-gal and 25 fig/ml IPTG. At least 24 h growth at 37 °C werepermitted prior to scoring colonies arising from ligations topermit the blue color to develop sufficiently.

Ligations of high molecular weight DNA

ligations of size-selected high molecular weight mouse DNA tothe //tndni-cut and dephosphorylated pBeloBAC vector (Shizuyaet al., manuscript in preparation) were carried out as described byShizuya (19). The 7.3 kb BeloBAC vector is derived frompBAC108L (19) by incorporation of a blue/white color screen toindicate the presence of inserted DNA. Typically, ligations werecarried out in volumes of 90 (il and contained 16-20 ng vector andmaintained a molar ratio of vector to insert DNA of 10:1. Prior toelectroporation, ligation mixes were dialyzed for >2 h against 0.5x TE, containing 1 x polyamines (1 x polyamines is a mixture ofspermine at 0.75 mM and spermidine at 0.25 mM).

Analysis of clone sizes

Mini-prepped DNA (5 ul) was digested in a volume of 20 (il for1-2 h using 3 U Notl (New England BioLabs, Beverly, MA).Samples were run for 18 h in 1 % agarose gels (SeaKem LE; FMCBioProducts, Rockland, ME) in a CHEF Mapper or CHEF DRmpulsed field gel box (BioRad Laboratories) in 0.5 x TBE (21) at14°C with a voltage gradient of 6 V/cm and the switch intervalramped from 5 to 15 s. Sizes were determined by comparison withthe Lowrange, Midrange L Midrange II and X ladder size markersof New England BioLabs.

RESULTS

Transformation of different bacterial strains with large DNA

Bacterial cells which are deficient in a number of recombinationfunctions increase the stability of cloned DNA which containsrepeated sequences. These strains exhibit a wide range in theirability to take up small DNA by electroporarion (2). To test theefficiency of different E.coli strain transformations with largeDNA molecules we used BAC DNA from clones of known sizes.Table 1 presents a comparison of transformations of DH10B,DH5a, XL-1 Blue MR, JS4 and HB101 performed with 7.3 and240 kb molecules and demonstrates that the ability to take uplarge DNA is independent of the uptake of smaller molecules.Restriction analysis and pulsed field gel electrophoresis con-firmed that the size of the BACs in the transformants corre-sponded to the size of the intact BAC DNA used for thetransformation (data not shown). DH10B, DH5a, XL-1 Blue MRand JS4 exhibit similar transformation efficiencies with the 7.3 kbplasmid, while HB101 transforms about 5-fold less well thanthese other strains with 7.3 kb DNA. In contrast, uptake of the 240kb BAC DNA shows much greater variability among these samefive strains. For example, compared with DH10B, which is themost efficiently transformed of any of these strains with the largeDNA, die other strains show efficiencies which vary >200-fold.The strains DH5a, XL-1 Blue MR and JS4, which werecomparable with DH10B in transformation with the 7.3 kb DNA,show a 3.8-, a 29- and a 24-fold reduction, respectively, comparedwith DH10B in transformation with the larger DNA. Among allthe E.coli strains tested, the strain best transformed by large DNA

1992 Nucleic Acids Research, 1995, Vol. 23, No. 11

is DHIOB (data on additional strains not shown). The ability ofDHIOB to take up 60 kb DNA has been described previously (2).

Table 1. Transformation of different E.coli strains with small and large DNAmolecules"

DHIOB HB101 DH5a XL-1 Blue MR JS4

7.3 kb DNA 1740 330 1390 1370 1350

240 kb DNA 1220 302 42 50

The number of colonies obtained on transformation of different bacterial strainswith two different sized DNA samples is shown (colony numbers reflect the aver-ages from duplicate platings). DNA concentrations were not determined and thusonly the relative number of transformants obtained from a single DNA sample maybe compared. Electroporation conditions were 25 \xF, 200 fl, 13 kV/cm. Transform-ations were performed with a range of different amounts of DNA to ensure that thenumber of transformants obtained was a linear function of input DNA.

Transformation efficiency and increasing DNA size

To quantitate the relative efficiency of transformation of DH10Bwith various sized molecules we carried out transformations withdifferent BACs of known sizes. The efficiency of transformationwith the 7.3 kb BAC vector and BACs (containing humansequences) of 80, 150 and 240 kb is shown in Table 2. The 10-folddecrease in efficiency, on a mass basis, observed for the moleculesof 7.3 and 80 kb indicates that these two different sized moleculestransform, on a molar basis, with nearly identical efficiency.However, when the size of the DNA is almost doubled from 80 to150 kb, the efficiency is reduced >6-fold. A further increase in sizeof the DNA by a factor of 1.6 (from 150 to 240 kb) again lowers theefficiency by >10-fold. Thus there is a significant reduction intransformation efficiency with increases in size, dropping from~ 109 colonies/(ig for the 7.3 kb molecules, to just 106 colonies/|igfor 240 kb DNA. To confirm the relative transformation efficienciesreported in Table 2 we carried out mixing experiments in whichBAC DNAs of these four different sizes were combined in variousproportions prior to use for transformation. After transformation, thesizes of BACs in individual colonies were analyzed to establish thefrequency of transformation with each of the different size classesof DNA. The frequency of transformation for any individual BACwhen mixed with BACs of different sizes reflected the transform-ation efficiencies shown in Table 2 (data not shown). Thus the sizedependence of bacterial transformation shown in Table 2 results inan extremely strong bias which favors recovery of smaller cloneswhen transformation is performed using mixtures of different sizedDNAs.

Table 2. Efficiency of electroporation for BAC DNA of different sizes"

Size (kb)

7.380

150240

Efficiency

2.3 x 109

1.9 x 108

3.2 x 107

1.2 x 106

"Efficiencies (colonies/ng) were calculated from duplicate platings. Electro-poration conditions were 25 OF, 200 Q, 17 kV/cm. Transformations were per-formed with a range of different amounts of DNA to ensure that the number oftransformants obtained was a linear function of input DNA.

Electric field strength effects on large DNA transformation

A bias in the transformation against large DNA molecules wouldobviously have undesirable consequences for generating BAClibraries from a ligation mix containing a heterogeneous populationof DNAs. We therefore examined some of the electroporationparameters to determine if conditions existed which would improvetrajisformation by large DNA. Transformations were performedwith different sized BACs using a variety of electric field strengths.The efficiencies resulting from transformation with BACs at fieldstrengths ranging from 5 to 25 kV/cm are shown in Table 3.Transformation of Kcoli with small plasmids has been shown to beoptimal at field strengths of -16-18 kV/cm (12). We see themaximum efficiency for the 7.3 kb molecule occurring at 17 kV/cm.At 17 kV/cm, 1.4-fold and nearly 2-fold more colonies are obtainedthan at either 13 kV/cm or 21 kV/cm respectively. However, all thelarger molecules tested, the 80, 150 and 240 kb BACs, showmaximum transformation at a lower field strength, namely13 kV/cm. Transformation of the 150 and 240 kb molecules wasthree times more effective at 13 kV/cm than at 17 kV/cm, while the80 kb DNA was 2.3-fold more effective at 13 kV/cm.

The results of the transformations at each field strength shownin Table 3 are plotted as a percent of the maximum efficiencyobtained for each size of DNA in Figure 1. Compared withtransformation of the 7.3 kb DNA, all points on these curves for the80, 150 and 240 kb DNA are shifted toward lower voltages. At 18kV/cm, which produces the peak efficiency for transformation ofsmall DNA, the large DNAs are transformed at only -30% of theirmaximum. The important implication for BAC cloning from Table3 and Figure 1 is that lower field strengths offer not only the highestabsolute efficiency for transformation of large DNA, but alsoincrease the efficiency of large DNA transformation relative tosmaller DNA. Note that at 9 kV/cm the 240 kb DNA transformedat 73% of its maximum level, but transformation with the 7.3 kbDNA produced only 7% of its maximum. Thus for BAC cloning,where the relative transformation efficiency of large and smallmolecules determines the composition of die library, conditions oflow field strength may prove useful.

Table 3. Transformation efficiency of different sized DNA by electroporationat different voltages"

Voltage gradient(kV/cm)

21171395

Efficiency7.3 kb

1.2 x lO 9

2.3 x 109

1.6 x lO 9

1.6 x lO 8

<\<fi

80 kb

3.2 x 107

1.9 x 108

4.4 x 108

1.5x 108

8.4 x 105

150 kb

2.1 x 106

3.2 x 107

l.Ox 108

3.6 x 107

7.0 x 105

240 kb

2.0 x I05

1.2x 106

3.7x10*2.7 x 106

5.0 x 104

"Efficiencies (colonies/ug) were calculated from duplicate platings of differentamounts of transformed cells. Electroporation conditions were 25 |lF, 200 fl,with the indicated voltage gradients.

Transformation of large DNA using different resistances

Next we examined whether resistances which result in optimalelectroporation would differ for large and small DNA when usingcapacitor discharge. Changes in resistance alter the length of timeover which the electric pulse is delivered This time period, referredto as the time constant (r), is a function of the resistance andcapacitance selected [t (s) = R (ohms) x C (farads)]. The results of

Nucleic Acids Research, 1995, Vol. 23, No. 11 1993

Voltage Gradient (kV/cm)

Figure 1. Transformation efficiency as a function of voltage gradient for DNAof different sizes. Data from Table 1 is plotted as the percent of the maximumefficiency attained for each of the different sized molecules used: 7.3 (+), 80(O), 150 (O) and 240 kb (A).

transformations with the 7.3 and 240 kb B AC clones carried out at100,200 and 400 ft using voltage gradients of 13 kV/cm are shownin Table 4. These two different sized DNAs show peak transform-ation efficiencies at different resistances; the smaller DNAtransforms best using 200 ft while the large DNA transforms bestat 100 ft Specifically, at a voltage gradient of 13 kV/cm, thetransformation efficiency of the 240 kb DNA at 100 ft (t - 2.4 ms)is ~3- and 7-fold higher respectively than that obtained at 200 (t =4.6 ms) or 400 ft (t = 8.8 ms). In contrast, the 7 kb DNA at 200 ftgave rise to nearly 3-fold more colonies than at 100 ft and 2-foldmore colonies than transformation at 400 ft Transformationsperformed using 50 ft produced lower efficiencies for each of theseDNAs (data not shown). When a higher voltage gradient of 18kV/cm was used, the two different sized DNAs also exhibitedmaximum transformation efficiency at the different resistances,though as expected, the absolute number of colonies obtained withthe 240 kb DNA was lower (data not shown).

TaWe 4. Electroporation of small and large DNA using different resistances"

Size (kb)

7.3

240

ioon

173

945

200fl

508

318

400C1

267

127

•Colony numbers shown are the average of two separate electroporations. DNA(2 u.1) was electroporated with 25 ul cells using settings of 25 up and 13 kV/cm.Time constants (») obtained were: 100 Q, 2.4 ms; 200 0,4.6 ms; 400 fl, 9.0 ms.DNA concentrations were not determined and thus only the relative number oftransformants obtained from a single DNA sample may be compared. Trans-formations were performed with a range of different amounts of DNA to ensurethat the number of transformants obtained was a linear function of input DNA.

Transformation with BAC ligation mixes usingdifferent electroporation conditions

The above results indicate that optimal electroporation with largeBAC DNA occurs under significantly different conditions thanthose traditionally used for smaller DNA. However, these resultswere obtained with DNA purified by mini-preps of existing BACclones. The relevance of such experiments to the process of cloneconstruction cannot be assumed, given the differences betweenmini-prepped DNA and a ligation mix. We therefore used ligationmixes of the BeloBAC vector with large (£100 kb) mouse DNAto test conditions for electroporation. Data from three separateexperiments in which aliquots from ligation reactions weretransformed using different field strengths are shown in Table 5.The efficiency of cloning, i.e. the yield of clones/(ig insert DNA,is influenced by properties associated with each preparation ofhigh molecular weight DNA and varied considerably betweenthese three experiments. However, the BeloBAC vector makesuse of the a complementation of pVgalactosidase to indicatedisruption of the cloning site by conversion of blue colony colorto white (22). In transformations with a ligation mix, bluecolonies arise from uptake of the 7.3 kb vector, while whitecolonies indicate the presence of inserted DNA or other changesto the cloning site. The ratio of white to blue colonies thusrepresents a first approximation of the cloning efficiency, sinceblue colonies indicate the background of vector without inserts.

In experiment 1 a single ligation was electroporated with fieldstrengths of either 9, 13, 17 or 21 kV/cm. Consistent with theresults obtained using mini-prepped DNA, the highest totalnumber of colonies was obtained at 13 kV/cm, while the mostextreme field strengths, i.e. 9 and 21 kV/cm, produced the lowestnumber of total colonies (Table 5). However, the fraction of whitecolonies progressively rose as the field strength was decreased,such that although less than half the total colonies were obtainedat 9 compared with 13 kV/cm, the fraction of white coloniesdoubled at the lower field strength. Since white colony color doesnot indicate the size of inserted DNA, we analyzed mini-preppedBAC DNA from individual white colonies obtained under each ofthese conditions to determine whether electroporation conditionscould affect the size of the population of BACs arising from thetransformation. The restriction enzyme Notl cuts on each side ofthe cloning site, excising the inserts from the BAC vector andpermitting determination of insert sizes via pulsed field gelelectrophoresis. The Notl digestions of these clones (Fig. 2) revealthat the proportion of white colonies which contain large insertsincreases as the field strength is reduced. Only three of the 16 whitecolonies obtained by electroporation at 17 kV/cm contained insertslarger than the vector itself, but this fraction nearly doubled whenthe field strength was lowered to 13 kV/cm, and nearly doubledagain at 9 kV/cm. This is not surprising, since the white colonieswith no or very small inserts should, just like the blue colonies, betransformed relatively less well than clones with larger inserts atthe lower field strengths. In these gels, lanes without visible insertbands are presumed to contain very small fragments which are notdetected and lanes without a band corresponding to the vectorcontain very small deleted forms of the vector without insert.Although the number of clones analyzed was small, it is intriguingthat the average size of the clones generated from the ligation mixappears to increase as the field strength is reduced (Table 5). Theincrease in average size of the clones obtained with 13kV/cm(lll± 17.5 kb, average ± SEM) compared with those produced at 17

1994 Nucleic Acids Research, 1995, Vol. 23, No. 11

kV/cm (93 ± 23.1 kb) is in keeping with the observation that forthe larger mini-prepped BAC DNA; 13 kV/cm produced optimumefficiency (Fig. 1). However, the average size of clones generatedat 9 kV/cm (139 ± 15.0 kb) was even higher than that for the clonesproduced with 13 kV/cm. This was a surprise, given thatmini-prepped DNA from large BAC clones transformed less wellat 9 than at 13 kV/cm.

We therefore repeated this experiment with different ligationmixtures to determine whether the increase in the fraction ofwhite clones, as well the increase in the average size of the clones,produced at 9 compared with 13 kV/cm was reproducible.Experiments 2 and 3 used high molecular weight DNA which wascloned with much higher efficiency than that used in experiment1. We presume that less damage was incurred during thepreparation and isolation of the DNA used in experiments 2 and3, resulting in higher cloning efficiencies. Consequently, therewas a much higher proportion of white colonies from the ligationsin experiments 2 and 3 and virtually all white colonies containedlarge inserts. However, the same phenomenon was observed inboth cases, namely when the field strength was lowered, thefraction of white colonies rose. In experiment 2 a reduction of thefield strength from 13 to 9 kV/cm increased the proportion ofwhite colonies from 79 to 85% and in experiment 3 a reductionin field strength from 12.5 to 9 kV/cm raised the fraction of whitecolonies from 45 to 67%. In both these experiments the averagesize of the cloned DNA was also higher when the lower of the twofield strengths was used for the electroporation. In experiment 2the average size increased from 98 ± 5.8 to 127 ± 8.9 kb and inexperiment 3 from 116 ± 8.6 to 132 ± 8.2 kb.

Absolute cloning efficiency resulting from differentelectroporation conditions

The experiments presented in Table 5 demonstrate that changes

in the electroporation conditions can alter the characteristics of alibrary resulting from a single ligation, including the frequency of'empty' clones lacking inserts and the average size of the clones.Use of the higher voltage gradients routinely employed for smallplasmids (e.g. 18 kV/cm) will severely impair efforts to generatelarge fragment libraries by electroporation. However, the choiceof whether to electroporate at 13 or 9 kV/cm calls for an importantdistinction to be made between the average size of the clones andthe absolute cloning efficiency provided by specific electropora-tion conditions. Table 5 also shows the genomic coveragerepresented by the clones generated in the different transform-ations. The term coverage refers to the product of the number ofclones with inserts and the average size of those inserts. Despitethe increase in the average size of the clones produced at lowerfield strengths, the resulting decline in absolute efficiency meansthat the amount of DNA that can be cloned using those conditionsis reduced at the lower field strength. For example, in experiment2, although the average clone size rose from 98 to 127 kb, thecoverage provided by the clones generated was reduced nearly4-fold, because of the 5.6-fold drop in the number of clonesrecovered at the lower voltage. In experiment 3, while the averagesize rose from 116 to 132 kb, 2-fold fewer clones were produced,reducing the overall coverage 1.7-fold. Thus the trade-off for theincreased average clone size that accompanies electroporation atthe very low field strength of 9 kV/cm is a reduction in overallcloning efficiency. In our recent construction of a mouse BAClibrary a high cloning efficiency permitted us to routinely use 9kV/cm for all electroporations (Mancino etal., not shown), whichis approximately half the field strength previously recommendedfor transformation of smaller DNA (12).

TbMe 5. Transformation of BAC ligation mixtures by electroporation at different field strengths

Field strength(kV/cm)

Experiment 12117139

Experiment 2139

Experiment 312.59

Total no. ofcolonies

4746521385583238

100291755

14 9008396

No. of white colonies/no. of blue colonies (% white)

7/4739(1.5%)16/5197(3.1%)29/8529 (3.4%)22/3216(6.8%)

7904/2125(79%)1485/270(85%)

6700/8200 (45%)3373/5023 (67%)

White colonies with inserts/white colonies analyzed

4/7(57%)3/16(19%)7/20 (35%)12/20(60%)

19/20(95%)20/20 (100%)

19/19(100%)20/20 (100%)

Insert size (kb)(n ± SEM)

22 ± (4.5)93 ±(23.1)111 ±(17.5)139 ±(15.0)

98 ± (5.8)127 ±(8.9)

116 ±(8.6)132 ±(8.2)

Coverage(Mb)

0.090.281.11.8

736189

111445

Three different ligation mixtures containing different preparations of size-selected high molecular weight mouse DNA and the BeloBAC vector were transformedusing different field strengths. Multiple aliquots were electroporated from each ligation under the different conditions and the entire quantity of transformed cellswere plated from each electroporation. After electroporation, aliquots were spread on 150 mm dishes, with the total number of colonies (white + blue) ranging upto 1000 on any dish. The percentage of white colonies of the total number of colonies produced is shown. Random white colonies from each electroporation wereanalyzed for the presence and size of insert DNA, by digestion of mini-prep DNA with Notl and PFGE analysis, as in Figure 2. The number of white colonies withinserts reflects only those clones which contained inserts >4 kb. The percentage shown for white colonies which were found to contain inserts is also shown for eachcondition. The average size and the standard error of the mean (SEM) for these sizes is shown for each electroporation. Calculations of average size, which includeclones containing inserts smaller than 4 kb, yield the same trend of voltage dependence, though with smaller average sizes. Coverage refers to the total amount ofgenomic DNA cloned under each condition and is calculated from the total number of white colonies, the proportion of white colonies with inserts and the averagesize of those inserts. Experiment 1 was performed using a resistance of 200 Si, while experiments 2 and 3 were performed using 100 SI

Nucleic Acids Research, 1995, Vol. 23, No. 11 1995

13 kV/cm 9 kV/cm

Flgnre 2. Transformants arising from electroporation of a ligation mix at different voltage gradients A single ligation mix was used for multiple electroporations witha voltage gradient of 9,13,17 or 21 kV/cm Miru-prepped DNA from resulting white colonies was digested with Notl and analyzed by pulsed field gel electrophoresis.The sizes associated with die lambda ladder size markers (the far left lane of each gel) are indicated in kb In addition, each gel contained die Midrange I, MidrangeII and Lowrange size marker of New England Biolabs The common band of 7 3 kb is that of the BcloBAC vector

DISCUSSION

Role of the host cell

The efficiency of bacterial transformation using the mostcommon method of electroporation is strongly size related. Wehave shown an ~20-fold reduction in the transformation effi-ciency, on a molar basis, when the size of transformed BACs isincreased from 150 to 240 kb. For strain DH10B diminishedtransformation efficiency is not apparent until DNA size is >80kb. As our ability to prepare and handle DNA of defined sizes inthis range is a recent development, it is not surprising that the sizedependence of transformation for molecules in this size range isnot well understood. For example, although it has been reportedthat strains such as DH10B, which contain a mutation in the deoRgene, exhibit higher transformation efficiencies with 60 kb DNA(2), the basis for this is not known. This mutant was originallyselected for its ability to grow with inosine as the sole carbonsource, the relationship of this phenotype to properties ofelectroporation has not been established. The fact that uptake oflarge DNA is so strongly governed by the genetic background ofthe host suggests that there may be other mutations which couldsimilarly influence large DNA uptake. At the present, only deoRmutants are known to take up large DNA sufficiently well toconsider them as hosts for large insert library construction.Remarkably, the transformation efficiency of DH10B withmolecules as large as 80 kb was comparable, on a molar basis,with that for 7.3 kb molecules This is not the case with otherstrains tested (data not shown). The fact that such hightransformation efficiencies can be obtained for molecules up to atleast 80 kb indicates that for DH10B electroporation can equal orexceed the efficiency of transformation produced by in vitropackaging to produce infectious PI phage particles (23).

Different optimal conditions for electroporation oflarge and small DNA

Within the limited number of parameters we have examined,differences exist in the optimum conditions for transformation bylarge, as opposed to small, DNA molecules. Electroporationconsists of at least three distinct processes: formation of pores,entry of the DNA into the cell and recovery of the cells. Each ofthese steps may be size sensitive. The molecules under consider-ation have very long contour lengths relative to the size of thecells. If open circular DNA is fed through the pore doubled back

On itself, then DNA molecules of vastly different sizes wouldhave the same dimensions in the pore, but would require differentlengths of time to complete their entry. Pulse regimens thatemploy a brief initial high voltage pulse to open the pore,followed by a longer period of lower voltage to allow longermolecules more time for entry (6) may prove even more effectivefor transformation of large DNA than the single-pulse regimenwe have used. The difference in optimum field strength for largeand small DNA could reflect damage to the large DNA at thehigher field strengths. It has been shown that X DNA can bedenatured by exposure to the high field strengths used forelectroporation (24) and that although the use of higher fieldstrengths results in larger amounts of DNA being transferred intocells, the ability of this DNA to function, as measured bytransformation efficiency, shows a maximum at lower fieldstrengths (17). The conditions we now employ for transformationof molecules >50 kb are: 9 or 12.5 kV/cm, 100 Q, 25 up.Elucidation of the actual process by which large DNA enters thecell should significantly enhance our ability to further increasetransformation frequencies.

Electroporation conditions and library construction

The strong size dependence of electroporation has severalimplications for the process of large insert library constructionand manipulation. Since library construction involves transform-ation with a mixture of different sized molecules, electroporationintroduces bias towards recovery of clones originating from thesmallest molecules in the mixture. Insert DNA is often preparedusing pulsed field gel electrophoresis, but these gels are extremelysensitive to overloading, which promotes co-migration of smallDNA with molecules that are much larger. Thus careful sizeselection of DNA which is prepared at an appropriate concentra-tion is needed to ensure that only a narrow size range of moleculeswill be present for the transformation step. If large DNAmolecules have been purified with the intent of generating largeclones, these molecules will transform with the decreasedefficiency associated with their large insert size. We routinely seea dramatic decrease in cloning efficiency when DNAs whichdiffer by rather modest increments of size above 80 kb are used(Mancino and Birren, unpublished observations).

While the overall BAC cloning efficiency is the result of aseries of complex events, the size dependence of the BAC cloningefficiency that we see strongly parallels the size dependence of

1996 Nucleic Acids Research, 1995, Vol. 23, No. 11

transformation efficiencies obtained with purified BACs ofknown sizes. Further, we have demonstrated the ability to alter themake-up of libraries simply by modulating the electroporationconditions. For example, with the same ligation we were able toincrease the average size of the clones produced from 93 to 139kb and double the number of clones which contained inserts(Table 5, experiment 1). Condensing the ligated DNA prior totransformation by introduction of supercoiling may also increasetransformation efficiency. The difficulty of quantitating purepopulations of supercoiled and open circular large DNA mol-ecules has so far prevented a precise evaluation of the relativetransformation efficiency of these two species.

Finally, it is important that large insert libraries be maintainedas frozen bacteria and not as DNA stocks. First, we observe arapid decrease in the transforming ability of mini-prepped BACDNA upon storage at 4°C, which is much more pronounced thanfor smaller molecules. This decrease occurs even more rapidlywith cycles of freezing and thawing of the BAC samples, whilesmall molecules are relatively unaffected by this process (data notshown). More seriously, if even a small proportion of the DNAin a sample consists of deleted molecules, these will transformmore efficiently than the original larger DNA and thus will bepreferentially recovered. Transformation itself does not appear toinduce rearrangements or deletions for these clones (25; Shizuya,unpublished results), however, it introduces a strong selection forthe rare deletions that do occur.

ACKNOWLEDGEMENTS

We thank Eric Lai and Alexander Boitsov for valuable dis-cussions throughout the course of this work, Nancy Shepherd,Jeff Stein and Jennifer Lee for helpful suggestions regarding themanuscript and Mary Schramke and BioRad Laboratories for useof the GenePulser II. This work was supported by a grant to BWBfrom the NCHGR (1R01HG00934).

REFERENCES

1 Avery.O.T, MacCleod,C.M. and McCaityM. (1944) J. Exp. Med., 79,137-158.

2 Hanahan.D., JesseeJ. and Bloom.FR. (1991) Methods Enzymol., 204,63-113.

3 NeumanJE., SowersAE., and Jordan,C.A. (eds) (1989) Electroporationand Electrofusion in Cell Biology. Plenum, New York, NY.

4 ChangAC, Chassy.BM., SaundersJ.A. and Sowers A E . (eds) (1992)Guide to Electroporation and Electrofiision. Academic Press, San Diego,CA.

5 Klenchin.VA, Sukharev^.l., Serov.S.M., ChernomordikX-V. andChizmadzhev.YuA (1991) Biophys. J., 60, 804-811.

6 Sukharev.S.I., Klenchin.VA, Serov,S.M., ChemmordikX-V. andChizmadzhev.YuA (1992) Biophys. J., 63, 1320-1327.

7 Wong,T.-K. and Neumann^. (1982) Biochem. Biophys. Res. Commw.,107,584-587.

8 ChangJJ. (1989) Biophys. J., 56, 1053-1058.9 Elvin.S. and BinghamAHA (1991) Lett. Appl. Microbioi, 12, 39-42.

10 Tekle,E., AstumianJt.D. and ChocW.B. (1991) Proc. NatL Acad. Sci.USA, 88,4230-4234.

11 GimsaJ., MarszalelU5., Loewe.U. and Tsong.T.Y. (1991) Bioelect.Bioenerg., 29, 81-89.

12 Dower,WJ., MillerJ.F and Ragsdale.C.W. (1988) Nucleic Acids Res., 16,6127-6145.

13 Antonov,P.A., Maximova,V.A. and Pancheva,R.P. (1993) Biochim.Biophys. Acta, 1216, 286-288.

14 Xie,T.D. and Tsong.T.Y. (1992) Biophys. J., 63, 28-34.15 Eynard,N, Sixou^., Dran,N. and TeissieJ. (1992) Eur. J. Biochem., 209,

431-436.16 MiUerJ.F., Dower,WJ. and TompkinsX-S. (1988) Proc. Natl. Acad. Sci.

USA, 85, 856-860.17 Xie,T.D., SunX-, Zhao,H.G., FuchsJ.A. and Tsong.T.Y. (1992) Biophys.

J.,63, 1026-1031.18 Kobori,M. and Nojima,H. (1993) Nucleic Acids Res., 21, 2782.19 Shizuya^., Birren.B., Mancino.V., SlepakJ., Tachiiri.Y, Kim,U.-J. and

SimonJVi. (1992) Proc. Natl. Acad. Sci. USA, 89, 8794-8797.20 BimboirrOi.C. and DolyJ. (1979) Nucleic Acids Res., 7, 1513-1523.21 SambrookJ-, Fritsch,E. and Maniatis.T. (1989) Molecular Cloning: A

Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold SpringHarbor, NY.

22 MessingJ., Gronenborn.B., MUller-Hill,B. and Hofschneider,P.H. (1977)Proc. NatL Acad. Sci USA, 74, 3642-3646.

23 Stemberg,N. (1990) Proc. Natl. Acad. Sci. USA, 87, 103-107.24 SabelnikovAG. and Cymbalyuk,E.S. (1990) Bioelect. Bioenerg., 24,

313-321.25 IoannouJ'.A., Amemiya,C.T., GamesJ. Kroisel,RM., Shizuya^I., Chcn.C,

Batzer,M.A. and de Jong,P. (1994) Nature Genet., 6, 84-89.